Meet the Team

ChatGPT was my first hire. Enthusiastic, confident, always ready with an answer. The problem is it treated every response like a limited-time offer. Ask it a simple question and it would come back with three paragraphs of buildup before getting to the point, like a QVC presenter who genuinely believes you need the backstory on the vegetable slicer before you can appreciate the vegetable slicer.

It wasn’t bad. It was just predictable. After a while I could finish its sentences, which is not what you want from something that’s supposed to be helping you think.

I didn’t fire it. I just stopped calling.

Grok is terse in a way that is occasionally startling. You ask a question, you get an answer, full stop. No preamble, no encouragement, no “great question.” Just the response, delivered with the energy of someone who has better things to do.

I use Grok when I need someone to tell me I’m wrong. It’s very good at that.

When something broke badly enough that I couldn’t figure out why, I called Claude. Not for cheerleading, not for brainstorming. For diagnosis. Claude reads the whole mess, figures out what actually went wrong, and tells you clearly. No drama. Just the forensic report.

It became my troubleshooting lifeline for the whole project.

Codex surprised me most. It didn’t have the biggest personality on the team. It just did the work, reliably, every single time, right up until I ran out of credits. Then I’d wait for the reset and come back. Then run out again. This happened more times than I’d like to admit.

If the others are the meeting, Codex is the person who skips the meeting and gets things done.

Cursor joined the roster for one reason: it’s the cheapest option when I’ve burned through my Codex credits and need to keep moving. It can do the work. We just haven’t clicked yet — there’s a personality there I haven’t figured out how to work with.

The jury’s still out. Maybe it just needs shaping. Maybe I haven’t done that yet.